Report says Google Gemini privacy controls can be hard to find

In short: A new report says Google’s Gemini AI is spreading across Gmail and other apps, and privacy and opt out settings can be confusing and hard to manage.

What's going on

Ars Technica reports that Gemini is showing up in more parts of Google’s products, including Gmail and Drive. Google says that when you use Gemini inside Google Workspace (work tools like Gmail and Drive), the AI can process your emails or files for a specific task, but it does not save that content or use it to train its main AI models.

The report says the situation gets less clear when Gemini can pull in information from your Google apps and then produce an answer that includes pieces of that information. Google says it can train Gemini on user inputs and outputs. In simple terms, your email is not directly “fed” into training, but a summary or snippet created from it could still end up in data used to improve the system, even if Google says it tries to remove personal details.

The report also describes opt out steps that are easy to miss. For example, turning off AI training requires changing a setting called “Gemini Apps Activity.” Doing that can also remove chat history, which forces a tradeoff between privacy and convenience.

What to watch

The report points to design choices that can discourage people from turning features off, sometimes called “dark patterns” (screens that nudge you toward what the company wants, like hiding the exit sign). It also says disabling Gemini in Gmail can require turning off “Smart Features,” which can also disable unrelated tools like inbox sorting tabs. Watch for whether Google makes Gemini controls easier to find in the main account privacy area, and whether regulators treat hard to find opt outs as a consumer protection issue.

Source: Arstechnica